Peter Blohm

I am a first-year PhD student at TU Wien, advised by Thomas Gärtner.

During summer 2026, I am a research intern at Aalto University in the group of Vikas Garg, where I previously worked on the composition and inference-time control of diffusion models (QuML).

Before beginning my PhD, I worked on interval-based embeddings for signed interaction graphs during a research internship at TU Wien, supervised by Stefan Neumann.

My PhD research studies how to support the safe deployment of complex machine learning systems through rigorous, data-informed guarantees. In particular, I am interested in certifying targeted and interpretable properties of black-box systems and using these certificates for deployment-time decisions such as monitoring, abstention, and control. My work draws on learning theory, verification, statistics, and probabilistic modeling, with applications to neural networks, generative models, and cyber-physical systems.

I also conduct applied research in medical data analysis at the Karl Landsteiner Institute for Outpatient Rehabilitation, focusing on predictive modeling and outcomes analysis in rehabilitation.

You can find more details in my publications.

news

Jul 17, 2026	Two workshop papers were accepted at ECML PKDD 2026: Iterative Indexing of Polycrystalline Laue Diffraction Patterns via Cluster-Driven Inference at ML4EMS and On the Best Interval Approximation Problem at MLG, the latter as an oral presentation.
Jun 15, 2026	I have returned to Aalto University for a summer research internship in Vikas Garg’s group.
May 15, 2026	I was recognized as a Gold Reviewer for ICML 2026.
Apr 30, 2026	Our paper was accepted to ICML: Temperature Scaling in Discrete Sequence (Language) Models — with Hannah Scheufele and Vikas Garg.
Feb 12, 2026	I started my PhD at TU Wien.

latest posts

Jul 17, 2026	Temperature Scaling in Discrete Sequence Models
Feb 13, 2026	Composition of Pretrained Diffusion Models

selected publications

ICML

Temperature Scaling in Discrete Sequence (Language) Models

Hannah Scheufele^*, Peter Blohm^*, and Vikas K. Garg

In International Conference on Machine Learning (ICML), 2026

ICML 2026 Poster

Abs PDF Blog Code Poster

Temperature scaling is widely used to control generation behavior and improve reasoning performance in discrete sequence models. However, sequence-level temperature scaling is challenging because the partition function is typically intractable and token-wise temperature scaling can bias maximum a posteriori estimates. This work introduces a unified formalism for autoregressive and diffusion-based language models, partition-free metrics for quantifying temperature scaling, and fine-tuning objectives that reliably achieve the desired change in model temperature. Experiments with GPT2 and BD3LM show more consistent generation with lower perplexity, with additional evidence for improved reasoning performance.
ICLR

Composition of Pretrained Diffusion Models: A Logic-Based Calculus

Peter Blohm and Vikas K. Garg

In International Conference on Learning Representations (ICLR), 2026

ICLR 2026 Poster

Abs PDF Blog Code

Composing pretrained diffusion models provides a cost-effective mechanism to encode constraints and unlock complex generative capabilities. Prior work relies on set-theoretic composition operators but suffers from limited mode coverage, biased sampling, instability under negation queries, and violations of core compositional laws. We introduce a principled calculus grounded in fuzzy logic, defining conjunction, disjunction, and negation operators that generalize classical mixtures and avoid these pathologies. Beyond existing methods, Dombi operators enable richer compositions such as Exclusive-OR, and we derive Feynman-Kac correctors to mitigate sampling bias. Empirical results on image generation with Stable Diffusion and multi-objective molecular generation support the conceptual, theoretical, and methodological benefits.
NeurIPS

Discovering Opinion Intervals from Conflicts in Signed Graphs

Peter Blohm^*, Florian Chen^*, Aristides Gionis, and 1 more author

In Advances in Neural Information Processing Systems (NeurIPS), 2025

NeurIPS 2025 Oral

Top 1.6% PDF News Code

Oral Presentation
ICML

Probably Approximately Global Robustness Certification

Peter Blohm, Patrick Indri, Thomas Gärtner, and 1 more author

In International Conference on Machine Learning (ICML), 2025

ICML 2025 Poster

Abs PDF Code

We propose probabilistic guarantees for adversarial robustness of classification algorithms. Traditional formal verification is often intractable for modern models, while sampling-only approaches lack formal guarantees. Our method samples an epsilon-net and evaluates a local robustness oracle on the sample, yielding probably approximately global robustness guarantees. The required sample size is independent of input dimensionality, number of classes, and learning algorithm, which enables application to larger neural networks than typical formal global verification methods. Experiments show that the approach better characterizes robustness than prior sampling-based baselines and scales favorably compared with formal methods.